I am getting acquainted with the OSM hot keys(keyboard shortcuts) and it makes the editing easier. At this point I have added over 800 buildings to my father’s village. It’s been fun so far and I’d like to move on to my mother’s village.
Happy mapping you all!!!
I’m currently working on it and to be honest I am unaware of what, and or how.
In most Yangtze Delta cities, buses usually operate in three separate networks: City Buses (城市公交), Suburban Buses (城乡公交) and Town-Village Buses (镇村公交). Some of those do operate outside the city boundaries, which are classified as Intercity Buses (毗邻公交), they can also be a City Bus Line, a Suburban Bus Line or a Town-Village Bus Line.
There exist a group of enthusiasts who spend their weekends from one city to another city purely by buses or subway lines, such a route is possible by benefiting from the tremendously complicated and connected bus network of Yangtze Delta Cities. I’m one of those enthusiasts who usually spend my entire weekend on buses and subway lines from city to city, as such I do exactly have “local knowledge” that I could benefit others if I can draw these lines on OpenStreetMap.
In such a circumstance I figured out that there isn’t a proper way to mark the type of a bus line. After a long discussion with some of the experienced users, I decided to add an unofficial tag network:route:bus:type to tackle this special issue around Yangtze Delta.
For values of network:route:bus:type I would temporary provide city, urban, suburban, town, intercity as five valid options, further extension is possible.
city, urban, suburban, town, intercity
1. Route 宜兴城乡公交225 is a suburban line and do have a stop outside the city boundary limit, we set network:route:bus:type=suburban,intercity to it.
2. Route 常州公交34 and 无锡公交26 are city, suburban, town-village mixed line and have stops outside the city boundary limit, since both city only has a single bus network, we set network:route:bus:type=urban,intercity to them.
3. Route 江阴-常州 is a purely intercity line and operates at two different cities, as such we only set network:route:bus:type=intercity to it.
When applying network:route:bus:type to a route, public_transport:version must set as 2 to indicate PTv2 is used. Then set public_transport:v2ext as yes to explicitly declare that this line is using unofficial extended features. Only set network:route:bus:type to the proper value when both of the public_transport tag is set.
After hearing advises and suggestions from the comment section, this flag is now deprecated and being replaced. Further more information will be updated soon on a separated article.
I was involved in mobilisation and signing up new members to OSM and participated in a mapathon titled “#mapmyvillage”.
One of the things that note-viewer does is showing notes on a map. They are displayed as markers which we can click to find notes in the table. Map markers show us locations and statuses of notes, the rest of note details can be found in the table. But we can dig deeper into these details. From the last diary entry we know that note-viewer looks for links to osm elements inside note comments. When clicked these links display their linked element on the map. That shows us the element geometry, but now we may want to see more details about the element. The map is displayed using Leaflet library1 which supports adding popups to various items that are shown over the base layer. A popup seems like an appropriate place to show the element version, changeset, last editing user and tags.
What’s the best way for us to implement these popups? If we just bind the popup (layer.bindPopup()) to the element geometry on the map, the user would have to click the element for the popup to show up. Sometimes that’s not very convenient to do because of note markers covering the geometry. To make things easier we can open the popup right away (layer.openPopup()) when the element link is clicked. But that’s not the only thing that needs to happen at that moment, and we may find ourselves struggling a bit with Leaflet. Obviously the linked element may be outside the current map view and we need to pan and zoom to it (map.panTo(), map.flyTo() or others). Opening the popup could have done part of this job because by default the map pans to the opened popup. That still doesn’t take care of zooming. But you may think so what, just let the user zoom to the tip of the popup to find the element on the map.
Here comes up our next problem: when the popup is opened at a low zoom level, after zooming it may drift away from the intended place. Anyway we probably want to zoom first and then open the popup. It would have been nice if we could queue the popup opening after the zooming/panning is finished. We’ll have to do that by listening to animation end events. There are zoomend and moveend events, which ones do we need to listen to? Maybe the map is already at a correct zoom level or at max zoom and the element is small. Then the zooming is not going to happen. Will the zoomend event also not happen in which case waiting for it is useless? But maybe the panning is also not going to happen and waiting for moveend is useless too? Apparently the moveend event happens after every call like map.panTo() or map.flyTo(), so we’ll wait for this event and then open the popup.
We’re still not done because for some unclear reason Leaflet sometimes manages to open the popup at a wrong location even after the map finished moving and zooming. Normally we would attach the popup to the geometry with layer.bindPopup() where we can also set popup options. One of the options is autoPan which is on by default as mentioned above. When the popup is opened at a wrong place, the map view scrolls away from the element with this option on, not something we want. With this option off, the popup stays outside the view, not something we want either. Here’s what we can do to solve this problem and what note-viewer currently does:
Now the popup is going to be shown at the center of the element’s bounding box. This is not ideal because this center may be outside the element itself, but at least it’s not at some random location. That’s true unless the user manages to interrupt the map movement by grabbing the map before the animation stops, but they’ll have to do that on purpose and be quick.
Everything written here is concerned with Leaflet v1.7.1 and may not be exactly true for the recently released v1.8. ↩
We have now run three casual mapping parties in the “Social Mapping Sunday” series, and they are always a bit different but each one has been as fun as the last.
In Shenton Park, lots of brand new mappers showed up and were introduced to OSM for the first time. About 13 of us in total! Meeting in the park and walking to nearby features to map worked well.
We didn’t come prepared with specific material to introduce new mappers, and we ended up spending a lot of time installing apps and setting up user accounts, etc. This was not a problem per-se, because it needs to be done. But it would have been smoother if we had a link to send people to which helped people get set up with the basics.
In Mount Lawley, 6 of us returning mappers set ourselves some explicit goals, and got a lot of data added to the map. We set the goal of mapping all of the shops and their names, footpaths, benches, bins, bus stops, bike parking and trees. We started in the middle of the shopping strip with a small team on each side of the road, and worked all the way to one end. Even though we didn’t map the whole retail strip, the are we did cover, we covered completely, which is very satisfying.
This gif shows the multiple stages of editing that these events encourage: The first step is cleaning up existing data and tracing features from imagery. This is important because it makes it easier to add data during the survey. It is also the best time to make assumptions based on the imagery, because any mistakes will be found during the survey. The next step is adding all of the data during the survey. The final step is cleaning things up, removing a few duplicates, fixing the odd tag, and generally making things neater with more powerful editing tools.
I have followed the same basic process that I described in my last post, for organising these events, but a few adjustments should be helpful. I have started a page on the wiki to document the process (in the hope that it continues to run, even when I can’t organise it).
It is important to do whole-group activities before everyone heads of to do mapping. That means before we start mapping we should: 1. take a photo of everyone, 2. decide on the details for the next event, and 3. decide on how to reconvene after mapping, for those who will stick around. Then 4. decide on mapping goals.
I have cobbled together a script for creating the before-and-after gifs showcased in this post. The most important part, is the awesome rendering engine Map Machine, which takes .osm files as input, so I can render the same area with the same settings passing in the data from different points in time (that I saved out of JOSM). I then use Image Magick’s convert to put the frames together into a gif.
(In April, I was not as active as before with OSM(F) stuff, because there was water damage in my flat, which was very stressful & time consuming to solve)
Social Media: 🐦 twitter: @lalonde / 🐘 fediverse/mastodon microblog: @email@example.com (rss feed)
not them ↩
Hi Andy and DWG:
I did not take composing this response lightly. I
have always mapped assuming good faith and intelligence
with co-mappers, both which have seemingly diminished.
If I got a message about a changeset, I first
checked the author’s edit count (i.e. experience), and
whether they were locally mapping. You can visit my
message list to see that I went well out of my way to
address legitimate queries.
If the author has minimal edits and no local
presence, I will not encourage trolls. I especially
detest, as probably many others also do, bird-doggers
who have zero interest in the local mapping, and whose
total focus is on policing, often using external tools.
I do not know of any individuals who enjoy being
watched over their shoulders.
A perfect example of this is:
The author had unbelievably zero edits and brazenly
stated, “I’m sure that you’ve been asked several
times”, when in actuality, I never had been.
Another mapper joined the recent fray from nowhere.
It looks like his modus operandi is already recognised:
And another with zero edits:
Andy, as for the “asked more than once previously”,
whom are you referring to? I only saw that one message
a few months ago. You insinuated that there was a huge
hue and cry. A bunch of BS on some external blog does
not cut it.
As for the comments, the two machines I
predominately use for OSM do not have keyboards, so
even writing a few words is a chore. Additionally, it
impossible to edit those comments, even when they are
misplaced and incorrect. Thus, I cannot imagine those
comments having any real value except for the pseudo
As for the sources, I am not sure if you know, but
the imagery source used is automatically added to the
OSM changeset database. Additionally, I frequently
checked and used all the imagery available on my
editors, so each node or way could have had multiple
exclusive sources. If I used external imagery, such as
Sentinel2, I normally attributed it.
Until May 3, everything seemed to be fine. I had
been contentedly and diligently mapping the same way
for years, without any objection. I have over 12 000
edits, which probably consumed well over 10 000 hours
of my volunteered and dedicated time. I prided myself
on top quality mapping and mapping entirely new routes,
On May 3, one rude, obnoxious and ignorant
individual named “Herman Lee” instigated and propagated
this recent shitshow, and that should be readily
apparent to any competent person. Why did he or you
not respond to my queries when I asked why he sent 8
messages in less than 90 minutes, all containing large
inexplicable binaries? I still consider his actions
nefarious, especially considering a zero explanation.
Andy, you sent me a (DWG) message to respond, and
then when I went to do so, I found you had blocked my
account. I hope you can discern the ludicrous irony
there? Maybe you think that that most OSM dweebs are
spending most their time in their mother’s basements
dedicated to OSM? I already know I spent (wasted?)
too much OCD time on OSM, and my intermittent
connections and other real responsibilities have
prevented any more.
As for my guide lines and circles that I have been
using for many years, it became acceptable to me when I
noticed that OSM was riddled with millions of metres of
“Hires” ways. I initially and ironically was deleting
some of them, as I erroneously thought they were
blatant commercial promotion for jobs, such as “Yahoo
Hires” and “Bing Hires”, versus the actually intended
term ‘high resolution’.
When I mapped new features, especially railway
curves, I initially estimated the size and radius, and
then with updated imagery, sized and aligned them more
accurately. Maybe you did not know, but railway
curves have a constant radius, which is well mapped
using a template circle. The beauty was, if the
placement was off, I only had to quickly move the line
or circle and all the attached nodes moved with it.
Conversely, each node would need to be individually
moved, which would take substantial effort and time.
I also extensively used Sentinel2Explorer on new
railway projects. Occasionally, I temporarily added
the “highway=path” tag to make the way visible on
Sentinel2Explorer, which AFAIK, was the only publicly
accessible ‘real time’ and recent imagery. I could
then get visible overlay feedback on OSM ways inside
Sentinel2Explorer after the change propagated. Often,
clear images could take weeks or longer. With what I
did, there was no lasting visibility on publicly
available maps. After I was satisfied with my way’s
accuracy, I deleted the guidelines. Maximal mapping
accuracy was consistently my objective, not adhering to
a few valueless directives.
I would be interested if a more proficient
individual than I could suggest any better method to
map new ways without existing imagery.
My temporary additions contrasted to the
proliferation of millions of incorrectly imported hash
tag highway paths and other objects destined to persist
indefinitely and which are actually non-existent. You
seemed to have been able to delete most my ~thousand
remaining guidelines in three seconds, which indicates
it was not actually a problem. The fact that drawing
each circle guideline consumed 30 minutes or more of my
time was immaterial. I do not think DWG will be
addressing with the same vigour all the useless “hires”
lines and millions of errant “paths”.
My enjoyment value had been steady declining. A
year or two ago, my WWW browsers stopped functioning to
edit OSM. A month or two ago, my WWW browsers stopped
functioning to even view OSM. A side effect from this,
is I cannot even respond to emailed OSM links, as I
have to first obtain another computer to view OSM, for
which I first need to forward the emailed queries to.
Minimal appreciation and maximal policing along with
an autocratic approach to any differences has
culminated to where I think it is best that I take a
sustained, and possibly permanent, vacation from this
ctrl + shift + o: opens the “Download object” dialogue, allowing you to download by ID
A whilst in node drawing mode: toggles accurate drawing mode, allowing you to align to common angles and with previous points. You can also hold ctrl and hover over another way to draw parallel with it
ctrl + alt + v: paste copied object into the source position (i.e. where it was when you copied it). Useful when working with multiple layers
5: zoom to downloaded area
4: zoom to conflict
ctrl + h: display history
holding ctrl while drawing a way starts a new way from a point
shift + j: join overlapping areas
alt + shift + f: opens the filter dialogue allowing you to hide certain objects
I’ve recently got into learning some of the more hidden functions in JOSM, and I’ve discovered a lot of cool dialogues and functions that I didn’t know existed before. Here are some of my favourite keyboard shortcuts (make sure you’ve toggled expert mode on):
I’m sure there’s many more that I’ve missed!
I did some statistical evaluation around StreetComplete, and in this post I describe what I evaluated and what the results are.
(This post is more or less the english text version of this YouTube video, which is in german. There is also a german version of this post here.)
First of all, how did I evaluate this data, or rather, where did I get all the info I’m talking about here?
The basis was the StreetCompleteNumbers script that I wrote some time ago. It’s a Python script that can be used to find out the number of solved quests for a user. The script is also available on GitHub.
One can use this script very easily:
from StreetCompleteNumbers import StreetCompleteNumbers
Then we just had to find out who are the users whose StreetComplete numbers we want to download. I tried to develop a method that makes as small a number of requests to the OpenStreetMap servers as possible. So simply downloading the entire changeset history for every user I come across should be avoided.
I was using the daily replication diffs since October 2021 (so since half a year ago). I downloaded every diff file, and looked at every changeset that occurred in it. If a changeset contains the changeset tag StreetComplete:QuestType, I trigger a download and save of its numbers for that user. Also, the program remembers for which users it has already saved StreetComplete numbers, so that they are not downloaded twice for the same user.
So I ended up with StreetComplete numbers for 5284 users, which was my database. And so we came directly to the first limitation of this evaluation. Only users who solved at least one StreetComplete quest between October 2021 and March 2022 appear in this statistic.
Before we get to the statistics, it should be said that I do not want to name any users in this evaluation, so we will only see the countries of the top users.
So, let’s get to the statistics now:
To do this, I created a list myself of quests that were disabled by default and checked for each user in my database to see if they had at least one solution for at least one of those quests.
The evaluation showed, one third of the users did not solve a single quest that was disabled by default. In theory, this doesn’t have to mean that these users haven’t activated any of these quests, but since many quests are deactivated because they are so spammy, I think we have a good estimate here.
The data I have indicates that over 14 million quests have been solved using StreetComplete. (To be precise, there were 13,975,938 quests at the time of the survey, which is why the percentages in the table below are based on this figure).
In StreetComplete, each quest has a color that assigns it to a rough category.
The categories are the following:
Here is the distribution as a graph:
And here as a table:
Here is a graph showing the 3 most frequently solved quests, as well as the quests that have been solved under 100 times overall by all users:
I think the quests with extremely few solutions can all be explained relatively easily:
Here is another graph that shows some more quests with their respective percentages:
And here is the whole list of quests in ascending order by the number of their solutions:
We can see very well that StreetComplete is quite undiversified in terms of its use and its users. Few people are responsible for the majority of all solved quests. Mainly StreetComplete is used to add details to houses, streets and roads. This doesn’t seem too surprising, since a city consists mostly of houses, streets and roads. In this respect, it is not a big deal that a large part of the solved quests are also related to these object types. As for the users, it looks a little different: Of the 5000+ users I examined, 100 are responsible for 30% of all quests solved. And while in the video version of this post I conclude that this is an indication of the not-so-good-looking diversification in OpenStreetMap, I’ve since done some more thinking and come to a different conclusion. You have to count out the power users much more than aligning the other users with them. So the average user doesn’t do proportionally little, but the power users do proportionally a lot. Looking at it this way, StreetComplete has a good user base that contributes relatively evenly to OpenStreetMap.
Just a question, when searching for any substation. As an example, here is Pembroke Substation:
However searching using the words “Pembroke Substation” returns significantly wrong results ie: in Bristol, Dublin, and even Florida!
The Name field is spelled correctly as “Pembroke Substation”, copy/pasting these exact words from the Name field into the search bar still can’t find it.
I have only been researching substations in South Wales so far, and they all have had good accurate naming, and are wither a Way if a single entity, or a Relation if more than one Way is used.
Is there a setting somewhere to enable this search? Maybe someone could drop me a message.
Many thanks, Scott
Annadruman Ring Fort, Castleblayney, Monaghan, Rep of Ireland
Hi, this is my first osm blog post :-)
Recently I stumbled upon some large areas that where mapped as forest, but Bing aerial imagery showed just a bleak, empty space. I was irritated first, shocked right after - having some thoughts about the earth climate in mind. Only some time later recognized: “Ok, wait. What do you think where your furniture comes from? This kind of thing has to happen. These are actually reneable resources.”
Album, comparing Bing, Esri and Esri Clarity imagery
So with less shock I went ahead to look around. I wanted to make sure that the imagery I was looking at, really was the latest one. The empty area and the visual progression between the different images already told a certain story. When zooming close into the Bing imagery, you could see the cut off stumps and occasionally even some stapled wood.
I tried to find dates for the different aerial imagery and also checked if there were any street level photos nearby. The Mapillary images I found were partially helpful. They were dated to October 2019 and there the stumps were also visible.
I had 3 different aerial image material available:
2. Esri Clarity
Bing material information did not show any dates for Bing, neither did Esri Clarity. But for Esri it mostly showed 22. Mai 2010, but partially 3. Mai 2017. You can clearly see the difference between existing trees in 2010 and cut off trees in 2017.
More to the south the Esri imagary was even from 2021, and there more cut off can be seen, which is not yet visible in Bing imagery (and also not yet mapped).
While it was interesting to analyze the imagery, my agenda with osm was a different one - to enhance the map. So for everybody waiting for the editing part, here you go. Let me jump into how I experinaced it and explain more considerations on the go.
My first thoughts were: hey there are no trees. If it is mapped as a forest, this is wrong. Double checking the wiki: ok mainly no trees, which means e.g. a clearing can still be part of the forest. But imho not a whole section that is gone.
Next question: what actually is a forest. I know the discussion between landuse=forest and natural=wood, but that’s not what I was after. Rather:
is it a larger, defined (maybe named) area of wood or can it be also parts of it (like a compound) or any (visual) collection of wood? - also answered in the wiki: Actually it could be any collection of (managed) wood, because for the defined/named parts there is boundary=forest (not often used) and boundary=forest_compartment.
Ok, so let’s go: Deleting all the wrongly tagged forests … … … I recognized soon that often I could not just delete them, because they where stretching partially over tree covered and no longer tree covered areas.
Only later I thought about if it was a good idea to completely delete these areas, as the geometry might have been interesting for nowerdays use as well, even when it is no longer covered with trees. However also I did not find an approriate tag for cut down wood. Maybe you can give your inputs / advice on this.
How did I deal with the overlapping geometry?
How did I do the split in iD editor? In case you know an easier way please let me know
Now coming to the point why this blog post is called “wood rotation” and not “deforestation”: When comparing different aerial imagery, to my surprise I found that in some areas, where nothing was visible on an imagery I rated older, in the newer imagery there were some green areas visible. First it made me overthink my hypotesis of image age. But soon I recognized that these had to be newly grown trees.
After understanding this it was pretty well visible by the form of the tree rows, that this trees had been artifically planeted to re-grow the forest (reforestation). So I looked for a matching tag and started mapping these (formerly unmapped) areas with landuse=plant_nursery. Small amount of them had existing mapping with natural=scrub which partially I kept where trees did not seem to be placed in structured rows, or changed otherwise.
Here are my changesets:
Is it in general a good idea to “micro”-map a forest that is constantly changing over the years? What should be ideally covered by a forest geometry? A visible part or a much bigger area (than can from time to time have wood or not).
While writing this blog I recognized the layer OS OpenMap Local with date April 2021. From what I understand this is a 3rd party open data map provider. If I interpret the green color correct, they are considering the whole bigger area (much more then visible on the screenshot) a forest, even when there is no tree cover everywhere.
You may want to mention that I have never been at this place nor asked people who have been there. That is correct. However I think that after the amount of considerations I did, I could be certain enough to do the mentioned updates. Of course I could have asked people earlier about their opinion, but while recognizing this in the middle-off I thought that I will just continue the changes with the most carefulness I can and hope that the positive aspects will predominate any potential negative ones.
In case you have more up to date information I would love to get in contact with you. As you might have seen there are more areas of tree rotation around that are awaiting to be mapped …
Even when I saw reforestation, I saw much more deforestation, which still makes me feel uncomfortable. If you want to influence the climate in a positive way, plant a tree :-)
To summarize, mapping this was very interesting and eye opening. I learned something about
Today I’m proud to present my new OpenStreetMap editor. It’s called Every Door and works on both iPhones and Androids. I shared the idea last Summer at a State of the Map, but started writing just late October. In the last month and a half thirty people made ten thousand edits with the editor and helped make it much better. Now I’m launching the open testing.
The official website has links to TestFlight and Google Play, a short video, and a FAQ.
I’ve got just one feeling: at long last. One way or another I was suggesting something like that for OSM since 2013. Made a failed attempt with OpenSurveyor. Watched with hope for big company projects with paid developers — but all these have disappeared. In this time we’ve got one amazing StreetComplete, which I like a lot, although it’s not for me.
Every Door is not StreetComplete. It’s not for everybody who’s got a minute to answer a few questions on the phone. This editor is for dedicated mappers. For her who, walking in a mall with hundreds of shops, thinks it’d be awesome to add each one of these to the map. For him who, pushing a stroller, dreams of mapping every sandpit and swing. For “fanatics” ready to add every street lamp, bench, and every tree in a park. This is a proper editor, with which you can forget about photomapping and trace recording.
Install Every Door today and help me make it better. The open testing will commence till late June at least, and in that time we’ll update the interface, polish input fields, add many heuristics inside. And on the way we’ll map hundreds of thousands of shops. Anything to make using the editor fun and exciting, so that our map grows even richer.
I am following ᚛ᚐᚋᚐᚅᚇᚐ᚜ as they are leading the way with a diary entry. This will be a work in progress entry until the end of May.
I plan to change some of the markup so that more information is referenced.
Now that the weather is more welcoming, i.e. dry, I’ve gone back to mapping walking and hiking trails in my area. I had done that before, either added the trail completely new as a relation with all that’s included, sometimes just added the trail markers, where the trail was already mapped. Most times, I try to do mapillary as well, sometimes just with the phone, sometimes with a 360° camera.
Today, I went to map the O’Gorman’s Lane Loop which is only a 4km walk (that is if you don’t get lost…), but it meets another, longer trail which was already mapped. But anyway, I ran into an American couple, Don and Kim who are exploring Ireland on rented motorbikes, but are also avid hikers and have hiked across the Czech Republic, Germany, Switzerland etc. Don was telling me that he has this app maps.me which shows all the trails and it’s free! And I said, well, it’s people like me who add those trails to OpenStreetMap which in turn adds it to your app. We parted ways, but I had a suspicion that I would run into them again after in the only café in the village. And I did. I walked up to them and gave him an openstreetmap.ie sticker (always handy in any bag I carry around). Once I had taken off my backpack, he could see the “OpenStreetMap surveyor” on my high viz vest, and the penny dropped. He had loads of questions and all the right ones, so I joined their table and chatted away with him.
He said that he wanted to map all the benches along the Camino. :D (We had noticed earlier that there are never enough benches along hiking trails in Ireland.) He also wants his name on one of those benches, and I explained that those little plaques can be added to OSM as well. He was very impressed, I think. They wanted to pay for my scone, but I had already paid. But we decided to share a taxi back, even though I would have been fine on the bus, and had told them about the bus as well.
He might look into it and become a mapper himself, who knows…It’s much easier to “convert” an OSM user into a mapper (even if they don’t know it’s OSM material), I would think.
PS: Videos about how to map hiking trails on my YouTube channel…
I started before COVID19. We had some success, but then COVID19 hit us and everything stopped. We are at it again.
Surveillance in Bengaluru
The Sahrawi Arab Democratic Republic does not exist in the annals of the United Nations, please remove it from the map.
Cartography is an artform. Besides being used for art, it is a practice rooted in subjectivity. For most people, it may be only a means of navigation, but any cartographer acknowledges that they make decisions based on their own opinions when making maps. OpenStreetMap is one of the more objective maps out there, but it’s still not always clear how to map things. As a community, we’ve had to make numerous decisions on the “best” way to map something. Sometimes, we don’t have a singular answer. Users of OpenStreetMap data must interpret these decisions as best they can. Tagging is usually what comes to mind when considering what comes into dispute here, but scope is important as well. And this is where privacy comes in.
We have some privacy standards. Besides GDPR compliance, the Data Working Group redacts edits that introduce personal information, such as annotations intended for an individual that may link their account to a real person. It generally isn’t acceptable to map features inside private residences either, like rooms or toilets. These may come as common sense to most, but others still could have counterpoints. The level of detail most are comfortable with is what is visible from street level or the sky. I think this is a good standard, but some are still left uncomfortable. There’s the occasional new mapper who deletes driveways leading to single-family homes. It may not even be their own driveway, but some may be unfamiliar with OpenStreetMap’s tagging system that makes it clear when a driveway is private and that it is, indeed, a driveway. One cannot fault a person for wanting privacy; the concern then is about damaging data (digital vandalism). At least where OpenStreetMap is based, there is no law against making a map of someone else’s property. Legal concerns about cartography are a separate matter not related to individual privacy, which is my focus here.
Does OpenStreetMap protect individual privacy? Depends on who you ask. I’d say so, but this essay isn’t about my opinions on privacy. It’s about ensuring people feel their privacy is respected by the site, within reason. Why? Because last year, I found someone who saw OpenStreetMap as a threat. So much so that they spent multiple days edit-warring, creating several accounts, and making vague legal-tinged threats at me for mapping their house and driveway. I was poorly equipped to deal with the user, and even though the Data Working Group was involved, the conflict only ended because the offending user gave up. My conduct began acceptable, but as I became exasperated turned pseudo-professional. It is important that users can deal with these issues before they are turned over to the Data Working Group, but I had nothing to go off.
I was confident I was in the right, but not all mappers might feel that way. Some might even take the vandal’s side. Because only I and the Data Working Group had any lasting involvement in the dispute, this issue has been forgotten. Nothing has been written about it. Only a brief discussion occurred in the OpenStreetMap US Slack server. As it happened over a year ago and was not recurring, I’d stopped paying it mind a while ago. I am sure that similar disputes exist, but I have not noticed them. I personally feel like I know what to do if I am involved in the following situation again, but it is just as important that others feel the same.
Enter June 2020. A user going by Hans Thompson has added AI-generated buildings in the outskirts of Anchorage, Alaska, using RapiD. Among these buildings is an unnotable house with a detached garage. The only tag on these buildings is building=yes.
February 2021 rolls around and someone signs up for an OpenStreetMap account under the username Privacy1. This user had previously created an account called Map_Manager that was used two years earlier to delete a litany of tracks and trails on private property. It is unclear if this was their own property. A week before creating this new account, they had returned to delete more trails from this property, but did not touch the house. One day before the new account was created, Hans also mapped the driveway leading to the house. This likely spurred the account creation, but no deletions were made until five days later. The edit was posted to Slack, prompting me to comment. My focus was made particularly on the changeset comment, which indicated to me that the user did not understand how OpenStreetMap functions. The user left a reply rehashing their changeset comment, leading to two other users explaining in other terms what I’d said.
Eight hours later, I reverted the changeset. Less than four hours later, they deleted the feature again, with the same comment. I requested they reply to my original comment, but to no avail. After just 20 minutes, I again reverted their deletions. Three hours passed and once again, they deleted the house, garage, and driveway, still with the same changeset comment. I left another comment notifying them that I had contacted the Data Working Group. At this point, I had quickened my pace, reverting their edit in just over seven minutes. At the same time, I started becoming irritated, and rather than use boilerplate about the changeset I reverted, I communicated my willingness to revert their vandalism as many times as was necessary. Four hours later, they returned, but this time with a new changeset comment, oddly accompanied by a French translation. This time, they pleaded for their privacy and to be left alone. Another user chimed in, explaining that their privacy was not threatened by the map features. At this point, I was asleep, so it took two hours, but I did come back. Refreshed, I switched back to boilerplate with my new reversion. They responded in equal time, but with a simpler message firmly requesting no “annotation” be made to their private property. Yet another user told them off, but still nothing was heard directly. Being a school day, it was nine hours before I could revert the changeset, but I persisted. They seemed to have caught on, as they set a new record with just a two-and-a-half-hour delay. No updates in the changeset comment department to report.. I left them another comment explaining what was wrong with their actions, but to no avail. I waited an hour for a vain attempt at submitting the 100 millionth changeset, but was slightly too late, and got number 100000019. Another new record was set by their hour and a half response time. One of the users that had told them off earlier in the dispute chimed in again, and I let him know the issue. I was a little passive-aggressive in expressing my disapproval that the Data Working Group had not yet taken action. This user reverted the deletion himself, which was greatly appreciated but in vain, as two hours later the user came back. Sick of writing changeset comments, an hour later I said nothing at all. They returned with a slightly revised changeset comment which I called out, but no one heard it. I decided to create a Wikidata entry for the home in the hopes that they would be unable to figure out how to delete it, but it did not help. I also used the city’s property information website to add the year the home was built. This mapping was done out of spite; call it silly, but it is a good act, or at least a neutral one. They had no more trouble deleting this, though. The aforementioned other editor again reverted the deletion, and added an explanation in the source tag of their changeset. It took eleven hours, but they did return, and deleted some other nearby features that weren’t even on their property to boot. We were clearly talking in circles at this point; I merely link every single changeset here to illustrate the user’s persistence. I laid out as clearly as I could in ten minutes that the user needed to stop in my next reversion comment, and that while I did respect their privacy, their privacy rights do not go as far as they claim. An hour and a half, a deletion, eight minutes, a reply quoting the dictionary at the user, a reversion, and then finally, a year long block. This was only a temporary roadblock (no pun intended) to them; 22 hours later, they returned with a new account they were clearly trying to pass off as a different person with a changeset comment insinuating the features in dispute were “mapping errors”, while talking only in French and using a French username. This one was blocked too, with a much more threatening comment. Perhaps the idea of legal action got them to wise up? A ten-year block finally scared them off. And so the features stand, with even more detail thanks to wonderful members of the community.
This subject has undoubtedly come up elsewhere, but there’s no guidance I could find on the wiki. I still have lingering questions, and here, now, I hope to pose them to other OpenStreetMap users so that we can all feel respected and be more welcoming to users with legitimate concerns. I’d love to hear your thoughts on the situation and these questions, or links to similar discussions.